Chandra and Wong studied the following discrete minimization problem:
Given a list of n positive real numbers, partition them into m parts s
o that Sigma q(i)(2) is minimized, where q(i) is the sum of all the nu
mbers in the ith part. They showed that the worst-case ratio of the LP
T rule (due to Graham), when applied to this minimization problem, has
a worst-case performance bound of 25/24. In this article we prove tig
hter bounds for this algorithm.